Corrective-RAG

An Implementation of this paper - https://arxiv.org/pdf/2401.15884 using LangChain and LangGraph.

Requirements

• Large Language Model (Change it in models/LLM. By default it uses gemini-1.5-flash by Google. You can use any model supported by langchain).
• Embeddings Model (Change it in models/EM. By default it uses text-embedding-004 by Google. You can use any model supported by langchain).
• Python 3.10 (use other versions at your own risk) (skip if using docker).
• Python libraries mentioned in requirements.txt (skip if using docker).

Setup

• Copy the repository.
• Change LLM and EM in models.
• Run the app -
  • Locally

    streamlit run streamlit_app.py
    

  • Docker

    docker build -t corrective-rag .
    docker run -p 8501:8501 corrective-rag
    

Check the app out here - https://crag.grecil.in/

To-do (contributions are welcome)

• Use Hybrid Search.
• Change VectorDB to Pinecone.
• Allow Users to Chat.
• Maintain Chat History for each user.
• Improve Theming.
• DOCX, XLSX, PPTX, CSV, TXT support.